tamasheq-99-2 / vocab.json
ad019el's picture
Upload tokenizer
532c863
raw
history blame
526 Bytes
{
"[PAD]": 43,
"[UNK]": 42,
"|": 0,
"ء": 12,
"آ": 36,
"أ": 34,
"ؤ": 40,
"إ": 16,
"ئ": 24,
"ا": 21,
"ب": 17,
"ة": 10,
"ت": 39,
"ث": 3,
"ج": 28,
"ح": 2,
"خ": 26,
"د": 5,
"ذ": 4,
"ر": 15,
"ز": 41,
"س": 30,
"ش": 20,
"ص": 19,
"ض": 13,
"ط": 23,
"ظ": 33,
"ع": 38,
"غ": 22,
"ف": 1,
"ق": 27,
"ك": 11,
"ل": 18,
"م": 29,
"ن": 7,
"ه": 6,
"و": 35,
"ى": 9,
"ي": 14,
"ٱ": 31,
"چ": 37,
"ڤ": 25,
"ک": 8,
"ی": 32
}